Infineon targets dense AI server power conversion
·

Infineon targets dense AI server power conversion

Infineon has launched a compact 120V dual-channel gate driver designed for high-density AI server power conversion and increasingly space-constrained multilevel architectures.

Infineon targets dense AI server power conversion
Summary
  • The 2EDL6014AC-G2D is a 120V dual-channel floating-output gate driver.
  • Its 2.2mm × 2.2mm package targets high-density switched-capacitor and multilevel power designs.
  • Infineon positions the component within a wider AI server power chain extending from grid input to processor-level conversion.

Infineon Technologies has launched a compact dual-channel gate driver for high-density power conversion in AI server designs.

The EiceDRIVER 2EDL6014AC-G2D is a 120V junction-isolated device with two independent floating output channels in a WQFN-12 package measuring 2.2mm × 2.2mm.

It supports dual high-side, dual low-side, and half-bridge configurations, allowing it to be used across switched-capacitor and multilevel power-conversion topologies.

Each channel can provide 4A source and 6A sink current.

Rack density creates component-level power pressure

Gate drivers sit far below the headline infrastructure layer of an AI data centre, but they influence how the MOSFETs inside high-frequency converters are switched.

That becomes more important as server designers try to deliver more electrical power through constrained board space.

Electricity is converted several times between a data centre’s incoming supply and the low voltages used by processors. Each stage creates losses, occupies physical space, and produces heat.

Higher switching density can reduce the size of parts of the power architecture, but tighter layouts also increase sensitivity to electrical noise and thermal performance.

Infineon has included independent INA and INB control pins and deglitch filtering for the two channels. The input stage is compatible with 3.3V and 5V logic.

Undervoltage lockout is provided for both floating output supplies, forcing the relevant output low when the supply voltage is insufficient.

An exposed thermal pad is intended to reduce thermal impedance and improve heat transfer from the package under sustained load.

From grid connection to the processor board

Infineon is positioning the device as part of a wider AI server power portfolio rather than as an isolated component.

The company’s products span solid-state transformers and circuit protection, server power supplies, battery backup, intermediate bus conversion, and later DC conversion stages.

It uses silicon, silicon carbide, and gallium nitride devices across that chain according to the switching, efficiency, voltage, and power-density requirements of each stage.

The 2EDL6014AC-G2D illustrates how the data centre power-density problem extends well beyond substations and UPS capacity.

As rack loads increase, power electronics inside the server have to move more energy through less physical space while limiting conversion losses. Those losses then become heat that must ultimately be removed by the facility cooling system.

Samples of the new device are available now. Its wider significance will depend on adoption by designers of server power supplies, intermediate-bus converters, and board-level power systems for the next generation of accelerated-compute hardware.


Stay updated with the latest insights and trends in the data centre industry by subscribing to our newsletter.

← Back

Thank you for your response. ✨